Cascading failures in the power grid, the spread of epidemics, bubbles in the stock market, the sudden popular rejection of a repressive political regime -- can these disparate phenomena, in so many different realms, have anything in common? Duncan J. Watts, a sociologist at Columbia University, thinks so. In Six Degrees: The Science of a Connected Age (Norton), he explains how the emerging science and mathematics of networks shed light on the behavior of connected systems. Assistant Editor Julian Sanchez spoke with Watts in February.
Q: Malcolm Gladwell's bestseller The Tipping Point suggested that a few well-connected early adopters can spark huge social changes. Is that right?
A: My interpretation is the opposite of Gladwell's. He has these "influentials" or opinion leaders, and you influence them, and they influence everything else. And it's just not true. We knew 50 years ago that this model was wrong. After the fact, and this is why Gladwell's book is so beguiling, you see that crime rates dropped or Hush Puppies took off and then you can always find the people with whom it started. But if it's something about them, why aren't they driving all the other trends?
What turns out to be the deciding factor is not the "influentials" but the people who are easily influenced. You might have someone who influences five times as many people as the average, but the total numbers relative to a population are still very small. Almost all of the action is away from the center.
We like the idea of influentials because then you can see cause and effect quite easily. What's actually happening is lots of people making decentralized decisions, but we want to be able to locate the reason for remarkable events.
Q: Why did Howard Dean's ballyhooed use of distributed social networks fall so flat?
A: Everybody thinks Dean was the "Internet candidate." But the Internet found Dean, not the other way around. Someone decided to start a Howard Dean Meetup, and it was only after that became popular that the campaign co-opted it and jumped on the bandwagon. It wasn't like they set things up that way.
And when things got tough, what did Dean do? He fired [campaign manager Joe] Trippi and said, "We want to centralize this organization." When no one is in charge, you have no idea what's going to happen. All the results that we have about how dynamical processes play out on networks show that they're tremendously unpredictable. Distributed systems are good at doing a lot of things, but it's hard to get them to do anything in particular.